Okay that requires modeling loss aversion which is considered 'irrational' and usually excluded from most (non-behavioral) economics, including this paper. Though of course not being rational doesn't make it any less real.

Probably one would want to include the degree of loss aversion in the model, for example, different classes of miners probably differ, but I won't pretend to suggest how.

In this paper increasing hash rate causes higher coin value/price due to greater perceived (and I suppose actual) security. I guess that is a similar result though via a different mechanism.